Both the neuropathology and our imaging data show that hippocampal formation (HIP) and neocortical abnormalities are found in the mild cognitive impairment (MCI) stage of Alzheimer's disease (AD). Much needed are improved biological markers that sensitive to progression and specific for AD pathology. Currently there are no known mechanisms accounting for disease progression. Our proposal is to use a novel MRI HIP imaging protocol and tested CSF biomarkers in order to: a) improve the prediction of cognitive decline, b) develop progression sensitive markers, and c) test a vascular mechanism for the progressive brain pathology of AD. We developed an MRI-Arterial Spin Labeling (ASL) method to give artifact free measures of cerebral blood flow (CBF) in the HIP and test the vasoreactivity to carbon dioxide (VR-CO2). Aim 1 is to evaluate among MCI patients, CBF and VR-CO2 as predictors of cognitive decline. Prior work shows that functional imaging modalities are superior to structural in the detection of AD related changes at the MCI stage. Our plan is to test the hypothesis that artifact free MRI measurement of HIP and cortical CBF and VR-CO2 are useful in assessing the future risk for cognitive decline related to AD, and perfusion imaging is superior to conventional volumetric methods. Because neither flow nor volume changes are specific for AD, a secondary goal is to examine the added contribution of two AD pathology-specific biomarkers: elevated CSF hyperphosphorylated tau (P-tau231) and decreased CSF amyloid beta 1-42 (A?42). Aim 2 is to test a candidate mechanism for AD progression. Both elevated plasma A?40 levels and intravascular A?40 deposits reduce HIP-CBF and VR-CO2 in transgenic mice with resultant parenchymal damage (volume loss) and inflammation. Our preliminary data show elevated plasma A?40 levels in MCI, and an association between elevated plasma A?40 levels and reduced HIP VR-CO2. We propose a longitudinal study of community residing elders to examine the hypothesis that regionally reduced HIP VR-CO2 predicts tissue volume loss and cognitive decline. Our plan is to conduct three clinical exams at 18-month intervals on 115 MCI (65-80 yrs) and 30 demographically matched normal controls. The primary outcome is decreased cognitive performance and the secondary outcome is clinical decline to AD. Five study hypotheses will be tested. Aim 1: H1) Baseline regional CBF and VR are useful for group classification and outcome prediction and contribute to volume measurement. H2) HIP CBF and VR increment CSF P-tau231 and A?42 in the prediction of outcome. H3) Longitudinal CBF reductions in AD vulnerable regions are superior to volume reductions in identifying patients with progressive cognitive impairments. Aim 2: H4) At baseline and longitudinally, HIP VR-CO2 is inversely related to the plasma A?40 level; and H5) Brain regions with reduced VR-CO2 will show progressive CBF and volume reductions. All the required clinical, laboratory, and imaging components for this study are standardized with quality controls. There are ample numbers of subjects available and adequate statistical power for hypothesis testing.